Microsoft has added branded-versus-non-branded query segmentation to the AI Citations dashboard in Microsoft Clarity, separating citations earned on queries that name a brand from citations earned on generic topic queries. PPC Land reported the release on 3 August 2026.
Microsoft is both the vendor and the owner of the product being measured here. Clarity is Microsoft's own analytics tool, the citation data is Microsoft's, and the sample figures published alongside the feature are Microsoft's illustration rather than a measured study. None of that makes the release uninteresting; it does mean the numbers should be read as a product demonstration.
The four things Microsoft's segmentation adds
According to PPC Land's 3 August 2026 report, the update has four components inside the AI Citations dashboard:
- branded labels marking individual queries in the queries card
- a Share of Authority breakdown split by branded and non-branded
- branded and non-branded filters applied across dashboard data
- citation analysis that separates brand-led demand from generic discovery
On the last of those, PPC Land wrote: "By separating brand-led demand from generic discovery, according to Microsoft, brand strength can be assessed more accurately." The attribution matters. That is Microsoft's claim about what its own segmentation achieves, reported as such.
Microsoft's five sample queries are an illustration, not a benchmark
Microsoft published a five-query sample built around a ski-retail example. These are illustrative figures from Microsoft's own demonstration, not measured benchmarks from a study, and they should not be quoted as industry averages or used to set a target. They are reproduced below because they show what the segmentation looks like in use.
| Sample query (Microsoft illustration) | Citations | Share of authority |
|---|---|---|
| best all-mountain skis (non-branded) | 2,500 | 20.3% |
| alpine beginner ski boots (branded) | 1,900 | 24.2% |
| alpine ski insulated snow paints (branded) | 1,600 | 22.3% |
| top snowboard brands (non-branded) | 1,100 | 18.6% |
| waterproof ski jackets (non-branded) | 986 | 15.9% |
The shape of the sample is the point rather than the values. The two branded rows carry fewer citations than the top non-branded row but a higher share of authority, which is exactly the pattern the split is designed to expose. Again, Microsoft built this example; nothing here was independently measured.
The third AI citation feature Microsoft has shipped in 25 days
PPC Land framed this as the third citation-related release in 25 days. Microsoft shipped Topic Insights on 9 July 2026 and a Query Topics beta on 22 July 2026, then this branded segmentation, reported on 3 August 2026. A vendor iterating on one dashboard three times in under a month is telling you where it thinks the demand is.
What Microsoft has not said
The gaps in this release are worth listing plainly, because several of them are the questions a buyer would ask first. The source does not state which AI assistants the citation data covers. It does not state whether the feature is free or paid. It does not describe the sampling methodology or rates, nor how the branded classification handles misspellings, sub-brands or competitor names appearing in a query. No conversion or business-impact metric is attached to any of it, and there is nothing about whether Thai-language queries are covered.
That last one is not a small omission for anyone working in a non-English market. A brand-name classifier is a language-dependent piece of engineering, and the source is silent on how, or whether, it handles Thai.
What this means for Thai marketers
The source does not address Thailand, so treat this section as reasoning rather than reported fact. The useful idea in the release is a distinction, not a number: being cited because someone typed your brand name and being cited because someone asked a generic commercial question are two different businesses.
A Thai brand with strong name recognition, the kind built over years of mass-media spend, can look healthy on total citation counts while winning nothing on generic queries. Every citation is coming from people who already knew the name. That is not visibility in an AI assistant; it is an existing audience checking a fact. Split the two apart and the picture changes, sometimes uncomfortably.
The non-branded side is where the acquisition work lives, and it is the harder half. It rewards content that answers the generic question well enough to be lifted into an answer, which is the same discipline behind visibility in AI-generated answers and the wider optimisation for AI search agenda. It also still rests on the fundamentals that make a page worth citing at all, which is why technical and content SEO has not stopped being the base layer.
Frequently asked questions
Which AI assistants does Clarity's citation data cover?
The source does not state it. PPC Land does not specify which assistants feed the AI Citations dashboard, so no claim should be made about coverage of any particular one. If assistant coverage matters to a reporting decision, that needs checking with Microsoft directly rather than inferring it.
Are the five numbers Microsoft published real benchmarks?
No. They are Microsoft's own illustrative sample data from a ski-retail example, not measured benchmarks and not a study. Treating a 20.3% share of authority as a target because it appeared in a Microsoft demonstration would be a mistake.
Is this feature free?
The source does not state whether the feature is free or paid, and there is no pricing information in the report. Anyone budgeting around it should confirm the commercial terms with Microsoft rather than assume.
Does the branded split work on Thai-language queries?
The source does not state it. Nothing in the report addresses language coverage, Thai or otherwise, which is a meaningful gap for a feature whose whole job is recognising brand names inside query text.
How does Clarity classify a query as branded?
The source does not describe the classification method. It does not say how misspellings, sub-brands or competitor names in a query are handled, which are exactly the edge cases that determine whether the branded and non-branded totals can be trusted for a given brand.
The short version
A dashboard that separates brand-led citations from generic ones is a genuinely useful distinction, and Microsoft shipping three citation features in under a month says the category is moving quickly. The caution is ordinary: the vendor measuring the citations owns the product, the published figures are a demonstration, and several basic questions about coverage and cost are unanswered in the source. If you want to know which half of your AI citations is actually winning new demand, that is a measurement question worth setting up properly before the tools settle.
